1551193 - Don't require re-creation of raptor-env each time a mozbase module is changed

Status: RESOLVED FIXED

(Testing :: Raptor, enhancement, P3)

Description

[Henrik Skupin [:whimboo][⌚️UTC+2]]

I was trying to run the tp6m-1 tests on my Android device and got the following error:

13:05:35 ERROR - Traceback (most recent call last):
13:05:35 INFO - File "/Users/henrik/code/gecko/testing/raptor/raptor/raptor.py", line 1202, in <module>
13:05:35 INFO - main()
13:05:35 INFO - File "/Users/henrik/code/gecko/testing/raptor/raptor/raptor.py", line 1165, in main
13:05:35 INFO - raptor.run_test(next_test, timeout=int(next_test['page_timeout']))
13:05:35 INFO - File "/Users/henrik/code/gecko/testing/raptor/raptor/raptor.py", line 952, in run_test
13:05:35 INFO - self.run_test_teardown()
13:05:35 INFO - File "/Users/henrik/code/gecko/testing/raptor/raptor/raptor.py", line 186, in run_test_teardown
13:05:35 INFO - self.playback.stop()
13:05:35 INFO - File "/Users/henrik/code/gecko/obj/debug/testing/raptor-venv/lib/python2.7/site-packages/mozproxy/backends/mitm.py", line 162, in stop
13:05:35 INFO - self.stop_mitmproxy_playback()
13:05:35 INFO - File "/Users/henrik/code/gecko/obj/debug/testing/raptor-venv/lib/python2.7/site-packages/mozproxy/backends/mitm.py", line 205, in stop_mitmproxy_playback
13:05:35 INFO - if self.mitmproxy_proc is None or self.mitmproxy_proc.poll() is not None:
13:05:35 INFO - File "/Users/henrik/code/gecko/obj/debug/testing/raptor-venv/lib/python2.7/site-packages/mozprocess/processhandler.py", line 858, in poll
13:05:35 INFO - raise RuntimeError("Process hasn't been started yet")
13:05:35 INFO - RuntimeError: Process hasn't been started yet
13:05:35 ERROR - Return code: 1

The problem is the following command line:

3:05:34 INFO - ['/Users/henrik/code/gecko/obj/debug/testing/mozproxy/mitmdump', '--replay-kill-extra', '--script', '""/Users/henrik/code/gecko/testing/raptor/raptor/playback/alternate-server-replay-2.0.2.py /Users/henrik/code/gecko/obj/debug/testing/mozproxy/android-amazon.mp""']

The --replay-kill-extra argument is only valid for 2.0.2, but not 4.0.4, and I had the 4.0.4 binary present in the folder:

$ ls /Users/henrik/code/gecko/obj/debug/testing/mozproxy
__MACOSX mitmproxy-4.0.4-osx.checksum
amazon.mp mitmproxy-4.0.4-osx.tar.gz
android-amazon.mp mitmproxy-recordings-raptor-tp6.checksum
facebook.mp mitmproxy-recordings-raptor-tp6.zip
google-search.mp mitmproxy-tp6m-amazon.checksum
mitmproxy mitmproxy-tp6m-amazon.zip

Somehow it is not downloading the 2.0.2 binary:

13:05:34 INFO - mozproxy downloading mitmproxy binary
13:05:34 INFO - mozproxy INFO - Cache matches, avoiding extracting in 'mitmproxy-2.0.2-osx'
13:05:34 INFO - mozproxy downloading mitmproxy pageset
13:05:34 INFO - mozproxy INFO - Cache matches, avoiding extracting in 'mitmproxy-tp6m-amazon'

To fix the problem I had to completely remove the folder $TOP_OBJ_DIR/testing/mozproxy.

I thought that this has been fixed by bug 1548574, but it doesn't look like.

Comment 1

[Florin Strugariu [:Bebe]]

From:
3:05:34 INFO - ['/Users/henrik/code/gecko/obj/debug/testing/mozproxy/mitmdump',

I see you are not using the latest version of mozproxy. in the latest version we create different folders for mitm2 and mitm4 and use the according path.

Can you update your raptor env or mozproxy install?

I know deleting the mitm folder fixed the issue. But if you switch back to the other version you will see this bug again.

Comment 2

[Henrik Skupin [:whimboo][⌚️UTC+2]]

It's unfortunate that you have to re-create the raptor-env each time a change has been made to any dependent mozbase module. We should really not pip install those dependencies, but take the code from the tree if possible.

I will just update this bug to have a starting point for for fixing the problem.

Comment 3

[Dave Hunt [:davehunt] [he/him] ⌚BST]

I believe this was fixed by bug 1564282.